import numpy as np
from sklearn.metrics import mean_squared_error
from ..tools.utils import einsum_correlation
[docs]def evaluate(reference: np.ndarray, prediction: np.ndarray, metric: str = "cosine") -> float:
"""Function to evaluate the vector field related reference quantities vs. that from vector field prediction.
Args:
reference: The reference quantity of the vector field (for example, simulated velocity vectors at each point or trajectory,
or estimated RNA velocity vector)
prediction: The predicted quantity of the vector field (for example, velocity vectors calculated based on reconstructed vector
field function at each point or trajectory, or reconstructed RNA velocity vector)
metric: The metric for benchmarking the vector field quantities after reconstruction.
Returns:
res: The score between the reference vs. reconstructed quantities based on the metric.
"""
if metric == "cosine":
true_normalized = reference / (np.linalg.norm(reference, axis=1).reshape(-1, 1) + 1e-20)
predict_normalized = prediction / (np.linalg.norm(prediction, axis=1).reshape(-1, 1) + 1e-20)
res = np.mean(true_normalized * predict_normalized) * prediction.shape[1]
elif metric == "rmse":
res = mean_squared_error(y_true=reference, y_pred=prediction)
return res